Manipulating Electoral Rules to Manufacture Single-Party Dominance
Kenneth Mori McElwain
Division of International, Comparative and Area Studies, Stanford University
[email protected]
July 2007
Keywords: Japan; institutional endogeneity; electoral system change; campaign regulations;
political parties; government longevity; incumbency advantage
Acknowledgements
I would like to thank Dan Okimoto, Anne Wren, Alberto Diaz-Cayeros, Beatriz Magaloni, Ethan
Scheiner, Rob Weiner, Barry Burden, Ray Christensen, Dick Katz, Jessica Weeks, and Catherine
Duggan for their advice at various stages of writing this paper.
the electoral data used here.
Steven Reed provided much of
Earlier versions were presented at the APSA Annual Conference
(2004), the Berkeley-Stanford Conference on Contemporary Japanese Politics (2004), and the
Japanese Politics Workshop at Harvard (2006).
Suggestions from the three anonymous
reviewers were critical in tightening the arguments in this paper.
All errors are my own.
Abstract
This paper argues that the LDP manufactured its parliamentary dominance in postwar Japan by
strategically altering specific facets of the electoral system.
More generally, I demonstrate that
intra-party politics play a crucial role in determining when and how electoral rules are changed.
Despite widespread evidence that the LDP would win more seats under an SMP electoral formula,
party leaders were repeatedly blocked from replacing the postwar MMD-SNTV system by
intra-party incumbents, who feared that such a change would harm their individual reelection
prospects.
However, party leaders had greater leeway in altering rules that generated fewer
intra-party conflicts.
Between 1960 and 1990, the LDP implemented approximately fifty changes
to campaign regulations, most of which were aimed at enhancing the incumbency advantage of all
rank-and-file MPs.
Statistical tests confirm that absent pro-incumbent revisions to the electoral
code, the LDP would have succumbed to declining public popularity and lost its majority at least a
decade earlier.
2
I.
Introduction
Winning more votes and seats is the sine qua non of government longevity, and many
scholars have analyzed how political parties manipulate regulatory and fiscal policies to increase
their electoral popularity.
This paper explores the strategic manipulation of an altogether
different set of tools – the electoral system.
Electoral rules are subject to partisan manipulation
precisely because they have non-neutral effects: larger parties generally prefer single-member
plurality (SMP) electoral formulas, while smaller competitors do better under a proportional
representation (PR) system.
By altering the electoral system into one that best matches their
vote distribution, political parties can maximize their electoral “bang for the buck” to win more
seats with the same number of votes.
Theoretical predictions of frequent electoral rule manipulation do not, however, have
strong empirical support.
Changes to the electoral system – especially before the 1990s – appear
to be rare in advanced democracies (Nohlen 1984), and the accepted wisdom in comparative
politics is that electoral rules are, for a variety of reasons, “sticky” (Pierson 2000).
Particularly
puzzling are cases where electoral rules remain untouched even though ruling parties can expect
to reap significant benefits from altering the institutional framework.
Using the empirical example of Japan between 1955 and 1993, this paper challenges two
common assumptions in the literature on electoral system endogeneity and offers a more nuanced
theory of how parties manipulate electoral rules.
First, major changes to the electoral system are
rare because of conflicts within political parties.
Altering the electoral formula or district
magnitude creates large-scale distributional asymmetries, such that even though the party as a
whole may benefit, not all rank-and-file MPs will be guaranteed reelection.
For the proposing
party to successfully alter the electoral framework, party leaders must use carrots and sticks to
convince backbenchers to support the legislation in a parliamentary vote.
However, not all
3
political parties are capable of enforcing internal compliance.
Leaders of the Liberal Democratic
Party (LDP) – a classically “weak” party – put forth three separate bills to replace the existing
MMD-SNTV formula during its single-party dominance between 1955 and 1993.
Even though
various simulations predicted that the LDP would win more seats under SMP, opposition from
rank-and-file LDP parliamentarians forced the withdrawal of each bill and left a suboptimal
electoral system intact.
Second, I argue that the literature serially undercounts the frequency of electoral reform
because it defines the electoral system very narrowly.
While “macro-level” changes to the
electoral formula and district magnitude do tend to be rare, political parties can successfully and
repeatedly manipulate more “micro-level” features of the electoral system, particularly campaign
finance and electioneering regulations.
Micro-level rules are more malleable because they
primarily discriminate between the campaign effectiveness of incumbents vs. challengers, and
thus can be designed to benefit all sitting MPs symmetrically.
The LDP – despite being unable
to change the electoral formula – altered campaign regulations forty-seven times between 1960
and 1990, and many of these revisions were targeted to strengthen incumbency advantage.
Specifically, the LDP increasingly restricted electioneering activities to a prescribed official
campaign period and then incrementally made this campaign period shorter, thereby minimizing
the ability of non-incumbents to attract voters.
Measures that improved reelection rates satisfied
the two principal intra-party actors: party leaders, who could keep their jobs as long as the LDP
maintained its majority, and party backbenchers, who supported any legislation that made it easier
to win their seats.
Despite being ignored by most studies of electoral rule change, this paper
shows that micro-level manipulation enabled the LDP to manufacture repeated electoral majorities
in the face of declining public popularity.
Japan under LDP single-party dominance is a useful case for analyzing electoral rule
4
manipulation for two reasons.
First, the LDP’s declining electoral margins generated strong
incentives to implement policies which increased their odds of staying in power.
Second, the
LDP’s single-party majority between 1955 and 1993 allows us to observe how intra-party
constraints may play a role in when and how electoral systems are altered, independent of
inter-party constraints inherent to coalition and minority governments.
This paper thus argues that political parties have the option of manipulating a diverse set
of electoral rules for partisan gain, but that the relative frequency of macro- vs. micro-level
reforms is a function of the ruling party’s ability to manage intra-party conflicts.
Section II
reviews the literature on political party decision-making, electoral system effects, and institutional
change more comprehensively, and discusses how the Japanese example fits into this broader
framework.
Section III analyzes changes to Japan’s Public Office Election Law and
demonstrates that the LDP implemented micro-level changes which weakened the electoral
prospects of opposition party candidates.
Section IV uses statistical tests to demonstrate that one
type of micro-level manipulation – the incremental reduction in the legal length of the campaign
period – significantly increased the reelection rates of LDP politicians.
Importantly, I find that
absent strategic electoral rule change, the LDP should have lost its majority at least a decade
earlier than the watershed 1993 election.
Section V concludes.
II.
Distributional Asymmetries from Electoral Rule Change
II.1
What we (think we) know about electoral rule endogeneity
Do political actors have strong incentives to manipulate electoral rules?
Theoretically,
any type of institution that has an asymmetrical impact on political outcomes should be unstable,
because empowered actors will try to modify it to suit their particular interests (Riker 1980).
The electoral system easily satisfies this criterion, as every electoral rule creates distributional
5
asymmetries in how seats are allocated between parties.
Studies have shown that differences
between first-past-the-post and proportional representation, the specific type of PR rule, and
perhaps most importantly, whether districts have single or multiple seats, all affect how party
votes are converted into parliamentary seats (Taagepera and Shugart 1989; Reed 1991).
The benefits to altering electoral rules are fairy straightforward.
Parties with larger vote
shares win disproportionately more seats under SMP than they do in PR systems (Taagepera and
Shugart 1989).
However, large parties that fear electoral slumps in the future, as well as parties
that are consistently small, prefer large-magnitude PR rules that minimize the chance of
significant losses (Shvetsova 2003).
As Sartori suggests, “[Electoral rules] are the most specific
manipulative instrument of politics” (1968, pg. 273).
Evidence is mixed on whether parties actually manipulate electoral rules to maximize
their seats and votes.
Many studies have documented that the initial choice in electoral rules –
following civil war, national independence, or democratization – is influenced by the strategic
calculus of the primary political actors (Boix 1999).
Similarly, country case studies find that
when rule changes do occur, they tend to reflect the preferences of parties-in-power (Bawn 1993;
Diaz-Cayeros and Magaloni 2001).
However, cross-national work on the frequency of electoral
rule changes indicates that electoral systems appear to be relatively stable over time (Lijphart
1994).
The accepted position in the literature is that electoral rules are inherently “sticky”.
Shepsle (1989) writes that institutional stability is due to high transaction costs that outweigh any
short-term advantages, while Pierson (2000) suggests that path dependency is at work, since
electoral rules are explicitly designed to resist change by having supermajority requirements for
revision.
As Gerard Alexander (2001) posits, however, neither the transaction cost nor path
dependency arguments are sufficient, since electoral rules do change periodically and, more
6
importantly, are subject to frequent attempts at revision.
A case in point is the resilience of the multi-member district, single non-transferable vote
(MMD-SNTV) formula in Japan during the Liberal Democratic Party’s reign between 1955 and
1993.
While the LDP held a single-party majority for close to four decades, it faced frequent
threats of electoral defeat.
The LDP’s vote share peaked in 1958 – the first election after its
formation – when it won 58%, but the party failed to win a majority after 1962.
telling, fewer people identified themselves as LDP partisans over time.
Perhaps more
According to monthly
polls by the Jiji Tsushin between 1960 and 1990, the LDP last had more than 50% public support
in December 1960, and more than 40% in January 1970.
The possibility of electoral defeat was not lost on the LDP’s leaders, who resorted to
various tactics to shore up the party’s declining popularity.
The government boosted public
works spending before elections to create the illusion of higher growth (Kohno and Nishizawa
1990) and gave policy benefits to key supporters during periods of economic crisis (Calder 1988).
However, the manipulation of fiscal and regulatory policy was at best a stop-gap solution,
especially after the 1980s when mounting public debt made irresponsible expenditures
problematic.
A straightforward way for the LDP to ensure a continuing legislative majority was to
alter the electoral system.
Between 1947 and 1993, Japanese elections were held in
medium-sized districts under SNTV, and various facets of this system helped the LDP maintain a
bare parliamentary majority.
It allowed smaller parties to win seats in larger districts,
diminishing incentives for opposition parties to coordinate candidates against the LDP
(Christensen 2000).
The high degree of malapportionment in favor of rural areas – the LDP’s
traditional support base – also helped the party garner more seats than warranted by their
declining vote share (Christensen and Johnson 1995).
7
MMD-SNTV was not, however, the optimal system for the LDP.
Indeed, there were
three high profile attempts by LDP leaders to alter the electoral formula: in 1956 under the
Hatoyama Cabinet, in 1973 under the Tanaka Cabinet, and in 1991 under the Kaifu Cabinet (Reed
and Thies 2001).
The Hatoyama plan (dubbed “Hato-mander”, combining “Hatoyama” with the
English term “gerrymander”) called for single-member plurality, while the latter two proposed a
mixed-member majoritarian system (MMM) with parallel single-member and proportional
representation districts.
Widely-cited contemporary simulations indicated that these disproportionate electoral
formulas would have generated a substantial seat bonus for the LDP.
The Asahi Newspaper – a
major national daily – predicted in 1973 that the LDP could have increased its seat share from
55.2% to 78.9% under MMM, and up to 89% should SMP be adopted (Hrebenar 1986, pg. 47).
Similar Asahi simulations in 1989 found that the LDP stood to win up to 375 seats – an
impressive 75% of the total – compared to the 59% that the party held at the time (Asahi
4/28/1990).
Despite such large, latent benefits under alternate electoral rules, the LDP never
succeeded in implementing institutional change.
In fact, only the first, Hato-mander bill was put
to a legislative vote; the two MMM plans of 1973 and 1991 were withdrawn before parliamentary
deliberations were concluded.
Why was the LDP unable to enact this obviously beneficial strategy?
Studies of
institutional endogeneity point to two crucial reasons why electoral rules might be stable.
First,
for electoral rules to change, they must be voted for through a defined legislative process, where
conflicts between political parties come into play.
The primary determinant of success in this
parliamentary stage is the distribution of legislative power, particularly the composition of
government.
In a single-party majority system, one party can essentially push through (or block)
8
institutional change as it sees fit.
In contrast, rule change is harder under a minority and/or a
coalition government, because the proposing party must negotiate with additional actors to get a
bill passed (Benoit 2004).
This logic fails to satisfactorily explain institutional stasis in Japan,
however, since the LDP did have a single-party majority when party leaders proposed electoral
reform.
Although consensus-building was the norm, the LDP had a history of steamrolling
legislation through the Diet when a time-sensitive bill was on the agenda and there was no chance
of getting other parties on board.
Examples include the renewal of the US-Japan Security Treaty
(1960), the adoption of a consumption tax (1989), and most relevant, the adoption of the
MMD-SNTV system itself in 1947.
A second set of factors contributing to institutional stasis is constitutional constraints.
Reform initiatives can succeed only if a group of parties has the necessary legislative fiat to
overcome any constitutional rules governing electoral rule change (Benoit 2004).
Constitutions
typically require higher hurdles for amendment than do legislative statutes – often a supermajority
in parliament, a national referendum, and/or an intervening election.
If electoral rules are
specified in the constitution, then changing them requires more than just a majority in the
legislature, rendering the system more resistant to partisan manipulation.
This constraint does
not apply to Japan, however, as the postwar Constitution is relatively lax, allowing electoral rules
to be changed at will by a parliamentary majority.i
Both parliamentary and constitutional restrictions are prominent factors in explaining the
stability of electoral rules generally, but they are by no means sufficient explanations, as
epitomized by the resilience of the MMD-SNTV system in Japan.
Instead, I argue that the
fundamental reason for institutional stasis is opposition within the LDP itself, manifesting as
conflicts between the party leaders and rank-and-file Diet members.
9
II.2
Intra-party constraints to electoral manipulation
The literature’s overestimation of the frequency of electoral reform derives from the
theoretical tendency to treat political parties as cohesive, atomistic units.
For electoral rule
change to be even proposed in parliament, however, a political party must first decide to put forth
a reform bill in the legislature, which ex ante requires internal consensus in favor of that proposal.
Indeed, we generally observe parties behaving cohesively precisely because they have resolved
internal conflicts over specific legislative provisions beforehand; bills that cannot garner
intra-party consensus rarely reach a parliamentary floor vote.
Intra-party negotiations can pit different sets of actors against one another, but the critical
disjuncture is between party leaders, whose goal is to maximize aggregate party seat share, and
rank-and-file politicians, who are primarily concerned with their own reelection – and only
secondarily with the performance of the party as a whole (Luebbert 1986).
Rank-and-file
members will reject any electoral rule change proposals which endanger their own survival; if a
sufficient number of them vote against a bill, then that proposal is unlikely to secure a
parliamentary majority.
Conflicts between party leaders and rank-and-file MPs arise when electoral reform
entails internal distributional asymmetries.
While a political party may win more seats in
aggregate under a different electoral system, this does not mean that all incumbents are
guaranteed reelection.
This problem is aggravated in the case of changes to macro-level features
of the electoral system – those that alter the translation of votes to seats.
Examples include the
district magnitude, which affects the level of proportionality within districts, and the electoral
formula, which establishes the method by which votes are aggregated and mathematically
converted into seats.
Large-scale rule changes are divisive because they are extremely blunt
weapons: although they may benefit the proposer party in aggregate, macro-level changes are
10
difficult to target to specific districts, and thus can catch some of the party’s own incumbents in
the crossfire.
The larger the scale of the change, the more incumbents will be affected, and
hence the greater the potential magnitude of distributional asymmetry.
The actual electoral costs from macro-level rule changes arise because some candidates
must move to different districts and develop a new personal vote base, which can weaken their
chances for reelection.
A significant portion of an MP’s incumbency advantage is based on the
reputation and network she has built up in her district over years of service, but these are ties that
do not necessarily carry over to new constituencies.
Looking at elections to the U.S. House of
Representatives, Desposato and Petrocik (2003) find that incumbents can lose 5 to 8% of their
vote base when redistricting incorporates a large number of new constituents.
Whether the rank-and-file can actually veto (or credibly threaten to veto) party proposals
depends on their degree of independence from the party leader.
If party leaders control valuable
resources that are crucial to reelection, such as campaign funds or cabinet portfolio, then
rank-and-file members may have no choice but to accept electoral rule change.
As others have
noted, however, intra-party conflict frequently prevented the LDP from implementing divisive
legislation, because party leaders lacked the coercive powers to enforce a party-line vote.
The
primary factor explaining rank-and-file autonomy is the electoral system itself: the traditional
MMD-SNTV electoral formula made the personal-vote more important than the party-vote (Carey
and Shugart 1995).
Candidates campaigned based on their own credentials, often highlighting
the amount of pork they returned to their districts, rather than on the programmatic profile of the
party (Ramseyer and Rosenbluth 1993).
Bawn, Cox, and Rosenbluth (1999) find that the vote
share of individual Japanese candidates are relatively unaffected by overall swings in their party’s
popularity, suggesting an electoral decoupling between individual and party performance.
The
primacy of candidate-based, local networks (koenkai) in electioneering made electoral rule
11
changes that entailed redistricting particularly costly for incumbent MPs.
Intra-party negotiation
thus became all the more critical, and LDP leaders were constrained from initiating electoral rule
changes that were likely to affect the reelection prospects of a significant portion of the
rank-and-file.
This paper’s contention, then, is that the type and frequency of electoral rule changes in
post-war Japan were determined largely by divisions within the LDP, rather than strategic
calculations with regard to other parties, as the existing literature tends to assume.
A striking
example is the breakdown in intra-party negotiations over adopting a mixed-member majoritarian
(MMM) system in 1991.
While opposition dissent complicated the passage of the bill, the
ultimate veto to Kaifu’s MMM plan (“Kai-mander”) came from the LDP’s own rank-and-file.
The most contentious issue was the creation of new single-member districts, which required a
significant reapportionment of seats and drew the ire of LDP incumbents.
Yoshinobu
Shimamura from the Watanabe faction argued that if the existing multi-member districts were
split into single-member districts, some incumbents would have to move to a different district
altogether.
He stated that “Leaving your district means death.
There are those who strongly
state that they will never leave their district, and will run as independents if necessary” (Asahi
1991/7/9; own translation).
One of the “rebel” leaders, Junichiro Koizumi, argued that SMDs
were not “temperamentally suited to the Japanese people” (Asahi, 1991/9/12).
Koizumi’s group
(Seiji Kaikaku Giin Renmei – “Federation of Diet Members for Political Reform”) claimed that
based on their own internal survey, 48 out of 97 respondents wanted to keep the MMD-SNTV
system.
With the very real possibility of internal revolt should the MMM proposal come to a
legislative vote, the Kai-mander attempt ended abruptly in September 1991, when the bill was
withdrawn from the Diet and Prime Minister Kaifu resigned.
As a caveat, macro-level electoral reforms did pass when there were countervailing
12
pressures that outweighed concerns over distributional asymmetries.
In many ways, these are
exceptions that prove the rule that intra-party consensus is difficult to achieve.
involves the reapportionment bills of 1964, 1975, 1986, and 1992.
One set of cases
Although the LDP, with a
heavily rural base, benefited from the rapid urbanization and resulting malapportionment of the
postwar period, the Supreme Court intervened in four instances by threatening to declare
malapportionment levels unconstitutional (Ohmiya 1992).
While the scope of proposed reforms
was relatively small, dealing with fewer than twenty seats in each instance, specific factions
mobilized en masse when districts with their incumbents were targeted for seat reductions.
In
the 1986 reapportionment debates, for example, the LDP’s Kawamoto faction – which had one
former and two current incumbents running in the six districts targeted for reductions – actually
supported an alternate plan by the Socialist Party, which proposed redrawing district boundaries
rather than changing seat apportionment.
Seat reductions were particularly problematic for
incumbents, who would have to compete against each other over fewer seats in a game of
electoral musical chairs.
In order to blunt internal opposition and get the reapportionment bill
passed, LDP leaders were forced to forge an agreement with the opposition Komeito, an urban
party which had no incumbents in the reduced-seat districts nor had any intention of nominating
new candidates there.
A second example of countervailing pressures is the successful electoral formula change
which occurred in 1993-94 – just two years after the failure of the Kai-mander proposal.
As
Reed and Thies (2001a) argue, politicians proactively take policy positions on institutional change
when they perceive voters as taking an interest, i.e. when there is an overwhelming political cost
to ignoring the issue.
In the failed 1991 Kai-mander case, public opinion was not in the
leadership’s favor; a series of polls showed that that over 70% of voters either disapproved, were
indifferent, or professed ignorance about reform (Reed and Thies 2001).
In contrast, the central
13
issue in the 1993 election was electoral system reform itself, driven by the emerging public
consensus (stoked by reform-minded politicians) that corruption was directly attributable to the
personalistic politics inherent to the MMD-SNTV system.
Forced to put forth competing
proposals of their own, LDP leaders dragged their feet and advocated either minimal change or
reforms that were disproportionately tilted in the LDP’s favor.
While this approach allowed the
LDP to assuage intra-party critics who had opposed reform in previous episodes, it led to massive
defections by electorally insecure, junior backbenchers and ideologically-driven, senior politicians
who sought to capitalize on the LDP’s hesitance and make their name as “reformers” (Reed and
Scheiner 2003).
The lesson, perhaps, is that party leaders cannot satisfy everybody all the time,
and that this condition can be critical for parties with weak internal discipline.
III.
The LDP’s Manipulation of Micro-Level Electoral Rules
The fact that Japanese backbenchers were relatively independent of the leadership does
not mean that the interests of the two sides never aligned in favor of institutional change.
Indeed,
incentives to alter electoral rules were shared by both groups when they affected all of the party’s
incumbents symmetrically.
Preferences are symmetrical – and hence intra-party agreement is
most likely – for what I call “micro-level” electoral rules.
These are regulations that influence
campaign effectiveness, or the ability of parties and candidates to 1) win over supporters, and 2)
convince those supporters to vote for them.
They include laws that define acceptable forms of
campaigning, the length of time parties can campaign, the amount of money that can be spent, and
penalties for violating these rules.
Micro- and macro-level rules affect elections in
fundamentally different ways: the former affects the capacity to win new votes, while the latter
deals with how those votes are converted into seats.
In terms of political impact, micro-level campaigning rules affect incumbents and
14
challengers differently, but they affect members within each group similarly.
This distinction is
crucial to the success of institutional manipulation, because everybody who votes for a statutory
or regulatory change is, by definition, an incumbent.
Accordingly, micro-level revisions that
benefit sitting MPs are more likely to receive unanimous support within parties, especially since
micro-level changes do not involve redistricting or seat apportionment – the very elements that
make macro-level changes contentious.
Ceteris paribus, most incumbents tend to prefer
restrictive campaigning regulations, because they are better known by virtue of their regular
legislative activities – and thus obtain lower marginal benefits from electioneering – than
non-incumbent candidates.
As such, parties with the most incumbents can consolidate their
majority by making micro-level rules more restrictive.
Figure 1 depicts the effect of micro- and
macro-level rules on the translation of popularity-to-votes and votes-to-seats, respectively.ii
[FIGURE 1 ABOUT HERE]
Japan’s regime of micro-level regulations is among the most stringent – and hence
pro-incumbent – in the democratic world.
The entire statutory handbook for Japan’s Public
Office Election Law, circa June 2004, is an astounding 2358 pages long (Senkyo-Seido
Kenkyuu-Kai 2004).
In contrast, the American “Federal Election Campaign Laws” (circa Feb
2005) span a sparse 234 pages.
POEL regulations include the number of posters allowed to be
distributed, the size of the posters, the number of speaking engagements permitted, the number of
politicians allowed to speak at each engagement, the number of cars that can be used for
campaigning purposes, the number of newspaper advertisements that parties can buy, and so on.
In contrast to the failure of macro-level electoral reform bills, over 40 changes to
campaigning regulations were made between 1960 and 1993.
For the purposes of understanding
15
the motives behind institutional change, the crucial question is whether these reforms
disproportionately benefited the LDP over the opposition.
With a majority in the Diet, any regulatory changes (or non-changes) that increased
reelection rates would allow the LDP to extend its dominance and benefit all of its backbenchers.
There are roughly two types of election regulation which enhance the incumbency advantage of
LDP Diet members.
First, the party benefits from increased restrictions on campaign activities
and media advertising.
When candidates have less time and flexibility to campaign, voter
choices turn increasingly on preexisting name recognition, which lies in the incumbent’s favor.
At the same time, the LDP has been significantly better at recruiting “quality candidates”,
particularly those with local government experience and established constituent networks
(Scheiner 2005), who can be electorally effective even in curtailed campaigns.
A second option
is to remove restrictions on campaign finance, especially on donations from private organizations
and corporations.
Since the LDP was perennially in government and maintained a stranglehold
over the policy-making process, it received significant support from the financial community,
construction companies, and agricultural cooperatives (Okimoto 1989).
This money could, in
turn, be rolled over into extra-electoral expenses, such as paying for local office staff and hot
spring trips for koenkai members.
In contrast, the LDP opposed public subsidies of elections,
since this would put the opposition, particularly minor parties such as the Democratic Socialists or
the Komeito (“Clean Government Party”), on a more equal financial footing with the LDP.
Does the pattern of POEL changes in Japan match the LDP’s preferences?
From an
epistemological perspective, the purposive manipulation of electoral systems is difficult to prove
definitively, since we must discern the private motives that underlie legislators’ public behavior.
There is, however, a preponderance of evidence that points to partisan intent in the timing and
content of electoral rule change.
First, the pattern of pro- vs. anti-incumbent regulatory changes
16
departs from more neutral recommendations made by a non-partisan electoral reform council.
Second, the number of micro-level changes increased as the proportion of beneficiaries – LDP
incumbents – increased.
Third, the exact content of new regulations demonstrates a consistent
bias in favor of LDP incumbents.
The following sections examine each of these points in greater
detail.
III.1
Ignoring recommendations for more open elections
Revisions to the Public Office Election Law, which governs all elections in Japan, were
rare until the late 1950s, when popular criticism mounted over the perceived dominance of
patronage politics and corruption.
The number of electoral law violations increased from 10,921
cases in the 1958 election to 17,177 in 1960 and 17,941 in 1963.
If we restrict these violations
to incidences of electoral bribery, the number of cases increased from 8,209 in 1958 to 13,346 in
1960 and 14,538 in 1963 (Shuugiin Chousa-kyoku 2002).
In order to review the causes of
money politics and restore public trust, Prime Minister Ikeda established the Electoral System
Deliberation Council (Senkyo-Seido Shingi-Kai) in 1961.
This Council was a non-partisan body
whose mission was to recommend both macro- and micro-level electoral rule changes which
would eliminate corruption and make elections “freer and fairer”.
Seven such Deliberation
Councils were eventually established between 1961 and 1970, each with a one-year mandate.iii
While the scope and magnitude of the Council’s reform proposals changed over time, the issues
covered – eliminating corruption and increasing fairness – remained relatively consistent
(Jichisho 1980).
The salience of the Electoral System Deliberation Council lies not in the fact that they
made many recommendations, but that much of the subsequent debate on micro-level rule
changes followed their guidelines.
For example, recommendations that dealt directly with
17
bribery and electoral violations were generally accepted with little debate, although not
necessarily in a timely manner.iv More contentious, however, were the Deliberation Councils’
anti-incumbent proposals to restrict campaign donations from corporate entities and deregulate the
range of campaigning activities allowed – both of which were anathema to the LDP.
In addressing these recommendations, the LDP’s primary tactic was to accept these
changes in principle, but dilute their actual content in the final legislation.
For example, the 5th
Council (1966) advised stringent cuts in private campaign funding, suggesting a ceiling of ¥10
million (approximately $91,000) on individual contributions, and ¥20 million ($182,000) on
corporate/union donations.v Subsequent Councils went further and argued that the reliance on
corporate donations produced perverse policy incentives for politicians, and that political parties
should be financed primarily by individual donations and membership dues. No significant
restrictions were enacted until 1975, however, when the Miki Cabinet passed a Political Financing
Reform Law (Seiji-Shikin Kaisei-Ho) that capped corporate donations at a much higher ¥150
million ($1.4 million), with a promise to revisit the issue in five years (a promise that was not kept
until 1993).
Even this bill was highly controversial, and only passed with a bare majority in the
House of Councillors by virtue of a tie-breaking vote by the Speaker.
While this seemed to be a
step in the right direction and somewhat assuaged public concerns, the only actual result was to
shift financing to other unregulated sources, such as private fund-raising parties (Cox and
Rosenbluth 1993).
Debates over deregulating campaign activities, such as lifting restrictions on door-to-door
campaigning and mass-media advertisements, are equally informative.
Door-to-door
campaigning was prohibited in 1950, but recommendations to remove this constraint were made
from the 3rd Council onwards.
Criticism of door-to-door campaigning came for various reasons:
it would lead to bribery; people could be intimidated into voting; Japanese people can’t say no
18
when directly approached.
Permitting door-to-door campaigning would have put the LDP at a
disadvantage, since it lacked the manpower to successfully compete in labor-intensive
electioneering compared to more grassroots-based parties, such as the Communists or the
Komeito.
At the 4th Council Meeting, however, representatives from the Justice Ministry and the
National Police Agency gave provisional approval, and many council members seemed resigned
to the fact that this ban was already impossible to enforce.
The LDP ignored calls for
deregulation, however, and the prohibition remains in the books to this day.
Indeed, Ramseyer
and Rosenbluth (1993) find that judges who supported overturning this ban were sanctioned by
the LDP, often finding themselves at the bottom of the promotion ladder.
Similarly, TV campaigning was only partially deregulated in 1969, when parties were
permitted to make limited policy-based speeches (seiken housou) on the national television station
(NHK), and again in 1983, when candidates were allowed to read out their resumes (literally) in a
staid newscast (keireki-housou).
In both cases, the number and length of TV appearances were
heavily regulated by the government, limiting the ability of challengers to champion their views
nationally.
Political parties were not allowed to purchase advertising time on commercial
broadcasts until the 1990s.
Table 1 lists all changes to the Public Office Election Law between 1950 (when the law
was established) and 2000.vi
Two points deserve particular notice.
First, a stark contrast can be
drawn between the origins of pro- and anti-incumbent revisions to the electoral code.
While all
anti-incumbent, deregulatory reforms were based on recommendations by the Deliberation
Council (italicized), most pro-incumbent changes were autonomous creations of the LDP.
We
can also observe that anti-incumbent reforms were initiated only after 1961, when the
Deliberation Councils began to recommend changes. Pro-incumbent revisions to the POEL, on
the other hand, were enacted from its inception in 1950, although the scope of changes increased
19
as the popularity of the LDP began to wane in the late 1960s and the need for incumbency
protection increased.
[TABLE 1 ABOUT HERE]
III.2
Increasing marginal utility of pro-incumbent changes
Two factors affect a party’s preferred magnitude of pro-incumbent bias in electioneering
rules.
First, the marginal benefit of tighter campaign rules is higher when the party has a large
number of incumbent candidates.
In this case, a governing party can stay in power by simply
holding onto the seats it already controls.
popularity of the party is low.
Second, tighter campaign rules are useful when the
If opinion polls leading up to the election exhibit declining public
support, then the party has poorer odds of winning new seats in contested districts.
As such, it
would gain more from defending current seats than trying to win new ones.
These two points explain why pro-incumbent biases to Japan’s Public Office Election
Law were adopted incrementally over thirty years, rather than implemented wholesale in 1955.
For the first decade of LDP control, the party’s public support ratings were fairly high and the
party’s ratio of incumbents-to-candidates was relatively low.
Armed with high ratings, the LDP
stood a good chance of defeating opposition incumbents and capturing new districts, while
holding onto their current seats.
The low incumbent-to-candidate ratio meant that the proportion
of LDP candidates who would benefit from tight campaign regulations was relatively small.
In
other words, initial demand for pro-incumbent rules within the LDP was low.
Over time, however, the LDP’s support ratings fell while the party’s ratio of incumbent
candidates rose to above 80%.
Figure 2 shows time-series electoral data between 1953 and 1993.
We see a clear increase in LDP incumbents and their victory rates over time, while the LDP’s
20
public opinion ratings in the month of the election mirror the trend downwards.
The falloff in
support ratings made it less likely that the LDP’s non-incumbents would win, while the higher
incumbent-to-candidate ratio increased the utility of – and intra-party demand for – pro-incumbent
electoral rules.
[FIGURE 2 ABOUT HERE]
Also interesting is the steady rise in the reelection rates of Japanese incumbents.
Something in the political environment – voter preferences, public policies, or electoral rules –
must have changed to improve incumbency advantage over time.
The following section teases
apart the precise content of POEL changes to demonstrate the creeping pro-incumbent bias.
III.3
Pro-incumbent changes as counteractive reforms
Given declining public support, the LDP could not completely ignore anti-incumbent
proposals by the Electoral System Deliberation Council.
The LDP was periodically plagued by
corruption and bribery scandals, most notably the Black Mist and Lockheed Scandals of the 1960s
and 1970s and the Recruit Cosmos and Sagawa Kyubin Scandals of the late 1980s-early 1990s.
By fixing perverse incentives in the POEL, such as prohibiting firms which received public funds
from giving donations to political parties, the LDP could stage a public mea culpa and halt the
steady decline in public support.
Eliminating the pro-incumbent bias of the POEL, however, posed a challenge to the
continuing electoral majority of the LDP.
To protect its single-party dominance, the LDP
counteracted anti-incumbent reforms with pro-incumbent changes that limited the former’s scope.
The most important change involved a separation of rules that applied before and after the start of
21
the official campaign period.
Under the POEL, politicians are generally prohibited from
canvassing voters in any way before the day elections are called (Kōji) – including asking
constituents to vote for them, distributing flyers and pamphlets, or placing newspaper
advertisements – but may do so in a limited fashion afterwards.
The LDP could diminish the
effects of electoral deregulation by first, allowing more flyers and posters to be distributed after
Kōji but not before, and second, shortening the actual campaign period itself.
As Table 1
indicates, the bulk of anti-incumbent, deregulatory changes between 1955 and 1993 applied after
campaigning began, leaving pre-campaign restrictions disproportionately stringent.
While all
major elections (both houses of parliament and prefectural leadership positions) started with 30
days of campaigning in 1950, the post-Kōji period for Lower House elections gradually dropped
to 20 days in 1958 and 15 in 1983, until by 1994, candidates could only campaign for 12 days.
In the same span, the campaign periods for Upper House and Gubernatorial elections fell to 17
days, and that for the Prefectural Assembly to 9 days.
LDP proposals to raise candidacy deposits, which can be reclaimed only if candidates
win more than a specified proportion of the votes, are another example of counter-reform.
Although the Electoral System Deliberation Council initially recommended an increase in
deposits to reduce frivolous candidates, the LDP took this suggestion to extremes.
While GDP
per capita only increased tenfold between 1950 and 1992, deposit rates for Lower House, Upper
House, and Gubernatorial elections increased a hundredfold, from ¥30,000 per candidate to
¥3,000,000.
If a political party wanted to run just one candidate in all 129 districts in 1992, it
would have had to pay ¥387,000,000 ($3.5 million).
Higher deposits make it difficult for
financially poorer candidates and parties to run in elections, thereby limiting opportunities for
entry by non-incumbents.
Because fluctuations in election results are correlated with the number
of fringe party entrants and spoiler candidates (Johnston et al. 1999), minimizing new competitors
22
allows incumbents to make more accurate predictions about the next election – and thus formulate
better strategies for how to win.
At the same time, higher deposits increased the electoral
salience of LDP factions, which leveraged financial aid to woo neophyte LDP candidates who
lacked personal wealth (Ramseyer and Rosenbluth 1993).
A simple analysis of the timing of micro-level changes confirms the counterbalancing
hypothesis.
First, anti-incumbent rule changes were more likely when government popularity
was declining, and hence the need for face-saving measures was greater.
The correlation
between the enactment of an anti-incumbent regulation in Diet session (t) and public opinion
swings in government disapproval from the last to current Diet sessions (t-1 to t) is 0.214.
Passage of pro-incumbent bills, on the other hand, is not significantly correlated with fluctuations
in government disapproval rates (0.093), but it is highly correlated with the enactment of an
anti-incumbent bill in the same Diet session (0.682).vii
The incrementalism in pro-incumbent biases also staved off public perceptions that the
LDP was drastically altering the electoral framework in its favor.
The average shift in Cabinet or
LDP approval rates in the month after pro-incumbent revisions is less than 0.5%.
contrast to the consternation raised by attempts to alter macro-level rules.
This is in stark
In 1973, when Prime
Minister Tanaka Kakuei proposed the adoption of a mixed-member electoral system, the
opposition parties immediately took to the streets (literally), disparaging the plan as
“Kaku-mander” and accusing the LDP of circumventing the will of the people.
On May 11th, the
Socialists, Communists, and Komeito publicly committed to stifling any debate on the measure
and threatened to block all 100 government bills that the LDP hoped to present during that Diet
session.
The three parties also began to distribute pamphlets criticizing the reform plan, and on
May 15th, orchestrated a public demonstration in 565 locations that mobilized 320,000 protestors
(Asahi Chronicle 2000).
23
IV.
Explaining the Effects of Micro-Level Rule Changes
While the level of symmetry in benefits determines the content and timing of micro-level
electoral rule changes, what remains to be explained is their substantive effect on election
outcomes.
Here, I examine the impact of gradual reductions in the official campaign period on
the probability of victory of individual candidates.
While the Japanese Public Office Election
Law is complex, a distinctive characteristic is the separation of campaigning tactics that are
permissible before and after Kōji, the day elections are called and when the official campaign
period begins.
One of the LDP’s most effective tactics was to reduce this campaign period over
time, making it more difficult for challengers to compete effectively against incumbents.
Notably, because the post-Kōji period was shortened gradually, we can operationalize it as a
continuous variable and measure the effect of each reduction on the election prospects of LDP
candidates.
This avoids methodological problems that complicate the analysis of other
micro-level rules, which were either never changed, such as the prohibition on door-to-door
campaigning, or were changed only once, such as restrictions on campaign donations.
IV.1
Specification of variables
To test the effect of reducing the campaign period on election outcomes, I use a logistic
regression model where the dependent variable is whether a given candidate wins a seat in her
district.
The dependent variable is coded as “1” when the candidate wins a seat and “0” when
the candidate loses.
The dataset is limited to candidates who have run in at least one previous
election to accommodate explanatory factors relating to prior performance; the time frame is
restricted to the LDP’s single-party majority between 1955 and 1990.viii
This yields a total
sample size of 9,375.
24
The independent variables include the factor of interest – length of the campaign period
– as well as other measures predicting election outcomes, such as party affiliation and district
characteristics. Kōji is a continuous variable that measures the number of official campaign
days leading up to the election (25 days = 1,873 cases; 20 days = 5,938 cases; 15 days = 1,564
cases).
Incumbent is a dichotomous variable that takes the value of “1” when the candidate is an
incumbent, i.e. had won a seat in the previous election (6,278 cases).
Incumbents should have a
higher chance of reelection based on greater name recognition as sitting parliamentarians, as well
as the policy tools they have at their disposal to win over additional voter support.
I also include
an interactive term, Incumbent*Kōji, to measure whether the length of the campaign period has
different effects for incumbents vs. challengers. Given my hypothesis that longer campaign
periods benefit challengers more than incumbents, the coefficient for Kōji should be positive,
while that for Incumbent*Kōji should be negative.
The model incorporates four party-level variables – LDP, JSP, JCP, and Independent –
which are dichotomous measures that take the value of “1” when the candidate is an official
nominee of the Liberal Democratic Party, the Japan Socialist Party, the Japan Communist Party,
or is unaffiliated (4,654, 1,863, 1,144, and 140 cases, respectively).
Since the LDP and JSP
were the dominant parliamentary groups of the postwar period, their candidates should have a
higher probability of winning, due to greater name recognition and funding.
The Communists,
on the other hand, have historically run at least one candidate in every district regardless of their
electoral prospects, and thus should have a lower likelihood of winning a seat as a group.
Similarly, political independents lack the financial resources or name recognition to compete
effectively, and thus should be more likely to lose. ix
An interaction term, LDP*Kōji is
incorporated to test whether LDP candidates do better when campaign activities are limited.
Much as incumbents benefit from shorter campaign periods, LDP candidates should do better
25
when opposition parties cannot criticize the LDP and sway the minds of voters effectively.
In
other words, while Incumbent*Kōji measures the effects of campaign period length on the
personal vote, LDP*Kōji measures its effect on swings in the party vote.
To take into account the electoral context of each district, I also include M, a discrete
variable signifying the number of seats in that district (median = 4; range = 2, 6). Ceteris
paribus, candidates have a better chance of winning when district magnitude is high, since the
proportion of votes needed to win a seat is lower.
Urbanization is a discrete variable where “1”
indicates the most rural and “4” signifies the most urban districts (median = 2).
While many
rural districts are personal bailiwicks of LDP (and sometimes JSP) candidates, urban areas have
historically housed more swing voters and thus enable a greater range of candidates to win
(Scheiner 2005).
Fixed effects for Year are also included to control for time-specific
fluctuations that affect all candidates.
To measure the past performance of each candidate, the model incorporates Previous
Vote Share, a continuous variable that measures the candidate’s vote share in the previous election
(median = 0.16; range = 0.00, 0.84).
Previous vote share establishes a baseline popularity level
for each candidate which should correlate with the number of votes – and hence the probability of
winning – in the current election.
While Incumbent captures this baseline popularity to some
degree, including Previous Vote Share disaggregates a pure incumbency advantage – namely
control over public policy at time (t) – from the factors that may have allowed that candidate to
win previously at time (t-1).
Finally, as a measure of overall party performance, I include a
continuous variable for GDP Growth, which is the change in GDP per capita from the previous
year to the current, as well as an interaction term, LDP*GDP Growth.
Ceteris paribus,
incumbent parties tend to do better when the economy is performing well (Lewis-Beck and
Stegmaier 2000); the prospects for LDP candidates should improve when GDP growth is high,
26
while non-LDP candidates should do better during periods of low growth.x
IV.2
Regression analysis
I analyze the effects of campaign period length on the probability of being elected using
two related logistic regression models, and the results are shown in Table 2.
The first model
incorporates party and district variables but leaves out Kōji and related interaction terms, while
the second model includes all independent variables.
Model 2 is obviously the more complete
model and includes our factor of interest, but the sign and statistical significance of the
independent variables remain consistent across specifications, signifying the robustness of the
models overall.
The ensuing discussion is based on Model 2.
[TABLE 2 ABOUT HERE]
The results mirror the stated expectation that shorter Kōji benefits LDP incumbents.
Unsurprisingly, the strongest predictor of winning is Previous Vote Share: the better the
candidate’s performance in the last election, the higher the probability of winning in the current
one.
The other coefficients are also in line with our intuitions.
Holding all variables at their
median values,xi an LDP Incumbent candidate wins 78% of the time, compared to 61% for
non-incumbents.
As for party affiliation, the victory rate of non-incumbent JSP candidates is
59% (almost identical to LDP non-incumbents), and for both Communist Party and Independent
candidates, a lowly 4%.
District characteristics also matter: candidates running in larger
districts have a higher probability of winning than candidates running in low M districts, while
the overall odds of winning are also higher in urban districts.
Similarly, GDP Growth has a
positive effect on LDP candidates but a negative effect on all other candidates, although the
27
substantive effects are small: a one standard deviation increase in GDP from five to nine percent
only increases the victory rate of LDP candidates by 0.1%
The key result, however, concerns the effects of reducing the official campaign period, as
captured by the variables Kōji, Incumbent*Kōji, and LDP*Kōji.
As hypothesized, challengers
generally do better when Kōji is longer, while incumbents have a higher chance of winning when
Kōji is shorter (as evinced by the negative coefficient on Incumbent*Kōji).
Similarly, LDP
candidates do worse when the campaign period is longer, when compared to opposition party
hopefuls (negative coefficient on LDP*Kōji).
To better depict how variance in campaign
lengths helped the LDP stave off challengers from opposition parties, Figure 3 plots the predicted
difference in the probabilities of victory between LDP incumbents and JSP challengers.
A larger
differential signals the relatively greater winning prospects of LDP incumbents.
LDP
incumbents were chosen as the baseline because their victory is what guarantees the LDP its
majority in the Diet.
The JSP, on the other hand, was the main opposition party prior to 1993,
and its electoral success would have contributed most directly to the LDP’s dethronement.
[FIGURE 3 ABOUT HERE]
Two separate values are calculated: (1) the predicted victory differential as the official
campaign period fluctuates in length, and (2) the predicted victory differential had the campaign
period remained constant at 25 days.
The shaded regions correspond to the area between the
upper and lower confidence bounds at the 95% level. Predicted values are calculated using
median values for each type of candidate in that given year (i.e. median LDP incumbent vs.
median JSP challenger characteristics in 1955, 1958, and so on).
Since 1955, the average
difference in the predicted victory rates of LDP incumbents and JSP challengers stays between 20
28
and 40%.
The simulation for victory differentials had the Kōji length always been 25 days,
however, illustrates the extent to which micro-level electoral manipulation helped the LDP.
If
Kōji had stayed at 25 days – the mandated period in 1952 – for the duration of the LDP’s tenure,
the victory differential would have been about 20% lower.
In both 1976 and 1983, JSP
challengers would have actually had a better chance of winning than LDP incumbents.
Figure 4 goes one step further and simulates LDP seat share had Kōji remained at 25
days.
Based on the logistic regression coefficients from Table 2, I calculate predicted
probabilities of victory for every candidate, and then re-rank candidates within each district in
order of that victory rate.
as new winners.
The top M vote-getters, where M = district magnitude, are classified
Two separate simulations are conducted: first, where only seats won by official
LDP candidates are tabulated, and second, where seats captured by LDP-affiliated independents
and candidates of the New Liberal Club – an LDP splinter party that nevertheless supported the
LDP in parliament – are also included.
The actual seat share of the LDP (including only official
LDP candidates) is depicted by the dotted line, while the simulated results are indicated by bars.
[FIGURE 4 ABOUT HERE]
Of the 4,903 LDP candidates whose victory rates are calculated, 3,320 who actually won
would have remained victorious even with Kōji at 25 days long, and 363 who actually lost would
still have lost.
However, the simulations indicate that 490 actual victors (10% of the total)
would have lost under a constant campaign period of 25 days.
As Figure 4 shows, this reversal
in victory is concentrated in two election years, 1980 and 1986, when the LDP would have won
46.2% and 48.2% of the total seats, respectively.xii
The LDP did, of course, actually win less
than a majority in 1976, 1979, and 1983, but in all three occasions, the LDP wooed conservative
29
independents into their fold after the election or ruled in conjunction with the New Liberal Club.
The simulations are less dire when including these two groups in the calculations, since the
LDP’s seat share increases by an average margin of three to five percent.
However, the
simulations also indicate that even with these bonus seats, the LDP should have lost its majority
in 1980, where the LDP en masse would have won 255 seats, or 49.8% of the total.
While the
LDP ultimately fell from grace in 1993, postwar Japanese politics may have reached its watershed
moment a decade earlier had the LDP not altered micro-level electoral rules strategically.
V.
Conclusion
The larger contributions of this paper are twofold.
First, it analyzes the endogeneity of
electoral rules comprehensively by expanding the definition of the electoral system to include
micro-level rules that are typically ignored in the literature.
At the same time, by comparing
incidences of both successful and failed electoral reform, this paper gets better traction of the
decision-making process within political parties.
Existing studies have long recognized that the
asymmetrical distribution of costs and benefits from electoral rule change create potential
conflicts between parties, but they have downplayed how the same asymmetries may pit members
of the same party against one another.
The first half of the paper shows that intra-party
consensus over macro-level changes is rare because benefits from reform are asymmetrically
distributed; the second half argues that this problem is mitigated in the case of micro-level
changes which generate symmetrical benefits.
Second, the paper demonstrates how difficult it has become for Japanese voters to “throw
out the rascals”, precisely because of the prohibitive electoral hurdles facing new candidates with
low name recognition.
While party affiliation matters for the electoral prospects of each
candidate, a greater distinction is whether she is an incumbent or a challenger.
For a party on the
30
wrong end of public approval fluctuations, tactics which protect incumbents are an effective
means for holding onto power, even if they reduce the electoral prospects of the party’s own
non-incumbent candidates.
The shortness of the POEL’s campaign period is of particular importance in
understanding the long-term electoral dynamics of Japanese politics.
Despite the promise of
programmatic politics under the new mixed-member majoritarian system of 1994, abbreviated
electioneering continues to make it easier for voters to evaluate retrospective performance than
prospective policy platforms.
This generates continued electoral pressures for candidates to
build up local support networks (koenkai), and this in turn benefits the LDP, which is better at
recruiting local politicians who already have an established vote base (Scheiner 2005).
In fact, most of the POEL’s restrictive regulations have remained intact despite the
fundamental change in the electoral formula. Prima facie, there have been some changes which
open up the election process, such as allowing privately-funded campaign commercials on TV,
granting public subsidies to parties to defray overhead costs, and lower ceilings on private
donations.
These reforms reflect the increasing clout of non-LDP parties after 1994, when
LDP-led coalition governments have become the norm. What the LDP giveth with one hand,
however, it quickly taketh away with the other: new rules since 1994 require any political posters
advertising candidate names to be torn down prior to Kōji (enacted 1999), limits the usage of
official campaign cars in distributing written pamphlets (enacted 2000), and most importantly,
reduced the Kōji period for Lower House elections to twelve days (enacted 1994).
Electoral rule
changes allowed the LDP to survive economic downturns, corruption and bribery scandals, and
declining demographic advantages, preserving its single-party dominance for thirty-eight years.
31
Bibliography
Alexander, Gerard. 2001. "Institutions, Path Dependence, and Democratic Consolidation."
Journal of Theoretical Politics 13 (3):249-70.
Asahi Chronicle. 2000. Asahi kuronikuru 20-seiki : Nihon to sekai no 100-nen Kanzenban ed. 10
vols. Vol. 5: Asahi Shinbunsha.
Bawn, Kathleen. 1993. "The Logic of Institutional Preferences: German Electoral Law as a Social
Choice Outcome." American Journal of Political Science 37 (4):965-89.
Bawn, Kathleen, Gary Cox, and Frances McCall Rosenbluth. 1999. "Measuring the Ties that
Bind: Electoral Cohesiveness in Four Democracies." In Elections in Japan, Korea, and
Taiwan under the Single Non-Transferable Vote: The Comparative Study of an Embedded
Institution, ed. B. Grofman, S.-C. Lee, E. A. Winckler and B. Woodall. Ann Arbor: The
University of Michigan Press.
Benoit, Kenneth. 2004. "Models of Electoral System Change." Electoral Studies 23:363-89.
Boix, Carles. 1999. "Setting the Rules of the Game: The Choice of Electoral Systems in Advanced
Democracies." American Political Science Review 93 (3).
Calder, Kent E. 1988. Crisis and Compensation: Public Policy and Political Stability in Japan,
1949-1986. Princeton, NJ: Princeton University Press.
Carey, John M., and Matthew Soberg Shugart. 1995. "Incentives to Cultivate a Personal Vote: A
Rank Ordering of Electoral Formulas." Electoral Studies 14 (4):417-39.
Christensen, Ray. 2000. Ending the LDP Hegemony: Party Cooperation in Japan. Honolulu:
University of Hawaii Press.
Christensen, Ray, and Paul E. Johnson. 1995. "Toward a Context-Rich Analysis of Electoral
Systems: The Japanese Example." American Journal of Political Science 39 (3):575-98.
Cox, Gary W., and Frances McCall Rosenbluth. 1993. "The Electoral Fortune of Legislative
32
Factions in Japan." American Political Science Review 87 (3):577-89.
Desposato, Scott W., and John R. Petrocik. 2003. "The Variable Incumbency Advantage: New
Voters, Redistricting, and the Personal Vote." American Journal of Political Science 47
(1):18-32.
Diaz-Cayeros, Alberto, and Beatriz Magaloni. 2001. "Party Dominance and the Logic of Electoral
Design in Mexico's Transition to Democracy." Journal of Theoretical Politics 13
(3):271-93.
Hrebenar, Ronald J. 1986. The Japanese Party System: From One-Party Rule to Coalition
Government. Boulder, CO: Westview Press.
Inoguchi, Takashi. 1986. "The Japanese Double Election of 6 July 1986." Electoral Studies 6
(1):63-9.
Jichisho, Senkyo-bu Senkyo-ka. 1980. "Senkyo-seido Shingi-kai Toushin/Houkoku-shu." Tokyo.
Johnston, R.J., David Rossiter, and Charles Pattie. 1999. "Integrating and decomposing the
sources of partisan bias: Brookes' method and the impact of redistricting in Great Britain."
Electoral Studies 18:367-78.
Kohno, Masaru, and Yoshitaka Nishizawa. 1990. "A Study of the Electoral Business Cycle in
Japan: Elections and Government Spending on Public Construction." Comparative Politics
22 (2):151-66.
Lewis-Beck, Michael, and Mary Stegmaier. 2000. "Economic Determinants of Electoral
Outcomes." Annual Review of Political Science 3:183-219.
Lijphart, Arend. 1994. Electoral Systems and Party Systems: A Study of Twenty-Seven
Democracies 1945-1990. Oxford: Oxford University Press.
Luebbert, Gregory M. 1986. Comparative Democracy: Policymaking and Governing Coalitions
in Europe and Israel. New York: Columbia University Press.
33
Nohlen, Dieter. 1984. "Changes and Choices in Electoral Systems." In Choosing an Electoral
System, ed. A. Lijphart and B. Grofman. New York: Praeger.
Ohmiya, Takeo. 1992. Senkyo-Seido to Giin-Teisuu no Zesei. 3rd ed. Tokyo: Hokuju Shuppan.
Okimoto, Daniel I. 1989. Between MITI and the Market: Japanese Industrial Policy for High
Technology. Stanford, CA.: Stanford University Press.
Pierson, Paul. 2000. "Increasing Returns, Path Dependence, and the Study of Politics." American
Political Science Review 94 (2):251-67.
Ramseyer, J. Mark, and Frances McCall Rosenbluth. 1993. Japan's Political Marketplace.
Cambridge, Mass.: Harvard University Press.
Reed, Steven R. 1991. "Structure and Behavior: Extending Duverger's Law to the Japanese Case."
British Journal of Political Science 29 (1):335-56.
Reed, Steven R., and Ethan Scheiner. 2003. "Electoral Incentives and Policy Preferences: Mixed
Motives Behind Party Defections in Japan." British Journal of Political Science
33:469-90.
Reed, Steven R., and Michael F. Thies. 2001. "The Causes of Electoral Reform in Japan." In
Mixed-Member Electoral Systems: The Best of Both Worlds?, ed. M. S. Shugart and M. P.
Wattenberg. New York: Oxford University Press.
Riker, William H. 1980. "Implications from the Disequilibrium of Majority Rule for the Study of
Institutions." American Political Science Review 774 (2):432-46.
Sartori, Giovanni. 1968. "Political Development and Political Engineering." Public Policy
17:261-98.
Scheiner, Ethan. 2005. Democracy without Competition in Japan: Opposition Failure in a
One-Party Dominant State. Cambridge: Cambridge University Press.
Senkyo-Seido Kenkyuu-Kai. 2004. Koshoku Senkyo Hourei-Shu. June 2004 ed. Tokyo: Dai-Ichi
34
Houki Shuppan.
Shepsle, Kenneth A. 1989. "Studying Institutions: Some Lessons from the Rational Choice
Approach." Journal of Theoretical Politics 1 (2):131-47.
Shuugiin Chousa-kyoku, Dai-Ni Tokubetsu Chousa-shitsu. 2002. "Senkyo-seido Kankei
Shiryou-shu." Tokyo.
Shvetsova, Olga. 2003. "Endogenous Selection of Institutions and Their Exogenous Effects."
Constitutional Political Economy 14:191-212.
Taagepera, Rein, and Matthew Soberg Shugart. 1989. Seats and Votes: The Effects and
Determinants of Electoral Systems. New Haven: Yale University Press.
35
Figure 1: How Electoral Rules Affect Election Outcomes
Party
Popularity
Vote Share
Micro-Level Rules
Electioneering regulations
Campaign spending
Political donations
Seat Share
Macro-Level Rules
Electoral formula
District magnitude
Threshold of representation
36
Table 1: Changes to Japan’s Public Office Election Law by Effect (1950-2000)
(Italicized if based on recommendations from the Electoral System Deliberation Council)
Year
1950
Pro-Incumbent
Abolish door-to-door campaigning
Decrease HR Kōji to 25 days; Increase
HR campaign deposits to ¥100,000
No free drinks/food to voters after Kōji
Limit opinion polls during campaign
period; Decrease HC Kōji to 25 days
Decrease HR Kōji to 20 days
Anti-Incumbent (Pre-Campaign Period)
Increase HR deposits to ¥150k; Decrease
HC Kōji to 23 days
Allow townhall meetings, speeches; Prohibit
donations from companies receiving public subsidies
1969
Increase HR deposits to ¥300k
Allow more townhall meetings/speeches
Allow more newspaper ads; Free TV time allocated
1975
Increase HR deposits to ¥1 million
Weak ceilings on corporate donations; Prohibit giving
money to constituents; Allow greeting cards
Allow more ads, posters
1981
1982
Restrict distribution of pamphlets
Increase HR deposits to ¥2 million
Reduce HR Kōji to 15 days; HC to 18
days; abolish joint campaign speeches
Prohibit cards, advertisements before Kōji
Increase deposits to ¥3 million; Reduce
HR Kōji to 14 days, HC to 17 days
Reduce HR Kōji to 12 days
1952
1954
1956
1958
1962
1963
1983
1989
1992
1994
1995
1997
1999
Abolish public funding of policy
advertisements by HC candidates,
organizations
No posters with candidate names prior to
campaign period
Anti-Incumbent (Post-Campaign Period)
Allow more posters, designate public poster boards;
Increase spending limits; Restrict campaigning by
civil servants
Allow more newspaper ads, posters; Establish more
poster boards
Establish more poster boards
Allow one extra televised speech
Parties eligible for public subsidies for all expenses
Greater limits to campaign donations
Increase campaign spending limits
Expand public funding for greeting cards, signs;
increase spending limits
Allow more private advertisements
Abolish corporate donations to candidates (OK to
parties)
2000
Expand public funding for advertisements
Data taken from Shuugiin Chousa-kyoku (2002); rule changes with minor or neutral partisan effects, such as increased penalties for electoral fraud or allowing
fishermen to vote by fax when at sea, are excluded from the table.
* Kōji = day elections are called and marks beginning of the campaign period; HR = House of Representatives; HC = House of Councillors (Upper House)
37
.2
.4
.6
.8
1
Figure 2: LDP's Candidate Profile and Electoral Success
1950
1960
1970
Candidates who are incumbents
LDP public opinion support
1980
1990
Incumbents who won
38
Table 2: The Effect of Campaign Length on Japanese Elections
Model: Logistic Regression, Robust Standard Errors
N = 9375
Constant
LDP
JSP
JCP
Independents
Incumbent
M
Urbanization
GDP Change
LDP*GDP Change
Previous Vote Share
Model 1:
Model 2:
Coefficient
Std. Er.
Coefficient
-6.681
0.391
0.680
-0.858
-1.192
0.006
0.857
0.076
-0.038
0.049
22.416
0.307***
0.120**
0.086***
0.108***
0.308***
0.077
0.043***
0.027**
0.015*
0.016**
0.966***
-8.895
2.622
0.679
-0.830
-1.585
0.840
0.852
0.077
-0.078
0.079
22.224
0.134
-0.042
-0.121
Kōji
Incumbent*Kōji
LDP*Kōji
Not displayed: fixed effects for “Year”
Pseudo-R
0.297
% Correctly Predicted
79.53%
Area under ROC Curve
0.821
*** p<.001; ** p<.01; * p<.05
2
Std. Er.
0.513***
0.396***
0.085***
0.107***
0.306***
0.367*
0.043***
0.027**
0.015***
0.017***
0.971***
0.020***
0.018*
0.019***
0.301
79.62%
0.825
39
Figure 3: Electoral Prospects of LDP Incumbents vs JSP Challengers
at Varying Campaign Period Lengths
15 Days
.1
.2
.3
20 Days
0
19
9
6
19
8
3
19
8
19
1979
80
6
19
7
2
19
7
7
9
19
6
19
6
3
19
6
8
0
19
6
19
5
5
19
5
3
0
19
5
Difference in Victory Rates
.4
25 Days
95% Confidence Intervals
Predicted at Actual Campaign Period Length
Predicted at Constant 25 Days Campaign Period
40
Figure 4: Predicting LDP Seat Share at Constant Campaign Period of 25 Days
70%
65%
LDP Seat %
60%
55%
50%
45%
40%
35%
30%
1958
1960
1963
1967
Predicted LDP Seat Share
1969
1972
1976
1979
1980
1983
1986
Predicted LDP + Affiliated Independents + NLC
1990
1993
Actual%
41
i
Article 47 of the Japanese Constitution writes, “Electoral districts, method of voting, and other
matters pertaining to the method of election of members of both Houses shall be fixed by law.”
ii
Like most dichotomies, the distinction between macro- and micro-level rules is fairly broad,
and there are alternate ways to separate out the effects of electoral rules which result in different
coalitions in favor of specific regulations.
For example, campaign finance rules that place caps
on private donations or distribute public subsidies for election expenditures create cleavages that
differentiate rich from poor parties.
Similarly, redistricting can benefit some LDP factions more
than others (generating intra-party cleavages) or pit urban against rural parties (generating
inter-party conflict).
However, the distinction I make between micro- and macro-level rules,
based on the capacity to win votes vs. seats, establishes a more basic differentiation between
those in and out of power.
Because micro-level rules engender fewer intra-party conflicts, they
provide the best forum in which to observe how the partisan motives of political parties drive
institutional manipulation.
iii
For the most part, the Council was successful in remaining neutral, although a small number
of politicians were always included along with academics, journalists, and other public thinkers.
iv
The suggestion to strengthen guilt-by-association rules (1st and 5th Councils; 1962, 1967),
whereby candidates could be penalized for illegal campaigning by their family members or
campaign managers, were adopted incrementally in 1975 and 1981.
A related proposal to
increase penalties for electoral fraud (4th and 6th Councils; 1967, 1970) was enacted in 1989.
v
Exchange rates conversions are based on figures from July 2005, when $1 ≈ ¥110.
vi
POEL changes in Table 1 also include revisions to the Political Funds Control Law (Seiji
Shikin Kisei Hō.
There were two major changes to the PFCL: in 1975, when private donations
42
to candidates were weakly capped, and in 1993, when public subsidies were granted to parties.
Minor changes in 1980 and 1995 altered reporting requirements for campaign donations.
vii
These intuitions are confirmed by a logistic regression model analyzing the probability that
pro-incumbent rule changes (the dependent variable) are enacted during a given parliamentary
session.
The regression restricts the data to the LDP’s heyday between 1960 and 1991, i.e.
beginning after the initial intra-party squabbles following the party’s formation and ending
before the factional defections which marked the end of the LDP’s single-party dominance.
I
control for fluctuations in public opinion support for the LDP, the Japan Socialist Party, and the
Cabinet more generally, as well as the number of months till the next election and the length of
the Diet session itself.
I find that the probability of pro-incumbent legislation increases by
44.9% in Diet sessions where an anti-incumbent bill was also passed.
In other words, the
strongest incentive to restrict electioneering regulations obtains when the LDP seeks to
counteract an anti-incumbent rule change.
Results from this logistic regression are available
from the author upon request.
viii
ix
Candidate and district data was provided by Steven Reed.
Independent does not include candidates who were tacitly supported by one of the parties but
were never official nominated.
Regression results do not change significantly even when these
“partisan independents” are included; results are available from the author upon request.
x
Models including other macroeconomic variables, such as inflation and unemployment, did not
affect the coefficient or standard error of other variables; they are excluded here to simplify the
analysis.
xi
M = 4; Urbanization = 2; Year = 1969; Legal Days = 20; Previous Vote Share = 0.155; LDP,
JSP = 0; GDP Change = 5.207
43
xii
The 1980 and 1986 elections were “Double Elections”, where Lower and Upper House
elections were held concurrently.
As Inoguchi (1986) writes, public interest in these elections
were stronger than usual, and the LDP benefited from the higher-than-expected turnout by
conservative voters.
44